Meta has released an early beta of Muse Code, a terminal-based AI coding agent designed to compete with Anthropic’s Claude Code and OpenAI’s Codex at a lower cost.
The new tool runs on Muse Spark 1.2, a coding-focused version of Meta’s latest AI model. Meta says the updated model improves code generation, complex debugging, codebase understanding, and complete developer workflows.
Meta Focuses on Lower Prices
Affordability appears to be one of Meta’s main strategies for attracting developers away from competing tools.
Muse Code uses the same standard pay-as-you-go pricing as Muse Spark. It costs $1.25 per million input tokens and $4.25 per million output tokens.
Meta will also offer a heavily discounted contributor tier. Users choosing this option must agree to provide feedback that Meta can use to improve the coding agent.
The contributor plan costs $0.10 per million input tokens and $0.20 per million output tokens. This makes the service significantly cheaper, although developers working with confidential or proprietary code may need to consider the feedback requirement carefully.
By comparison, Anthropic’s Sonnet 5 model normally costs $3 per million input tokens and $15 per million output tokens.
Muse Code Can Build and Test Software
Muse Code can complete software engineering tasks using text instructions. It can plan changes, write code, and check whether the finished results work correctly.
The agent can also coordinate several sub-agents and assign different parts of a project to them. This allows multiple tasks to run in parallel and could help it complete larger projects more efficiently.
Meta demonstrated Muse Code through several sample projects. These included an interactive model of a photon sphere, a game inspired by Plants vs. Zombies, and a website created by analysing an MP4 video.
The pricing difference between Meta and its rivals could encourage companies to choose Muse Code if its performance proves competitive. A similar shift has already benefited Chinese AI providers, as some businesses have adopted models such as DeepSeek to reduce the rising cost of using American AI services.
The post Meta Launches Cheaper Coding Agent to Punch OpenAI and Anthropic appeared first on ProPakistani.
